In the days leading up to caucus night, polls showed Mitt Romney as the likely winner in Iowa, though Rick Santorum was gaining steam and Ron Paul seemed poised for a strong showing.
By Wednesday morning, they were the top-three finishers respectively. But with only eight votes separating Romney and Santorum for first, the question remains: how accurate were the surveys of Iowa Republicans?
When it comes to horserace polling, presidential caucuses provide a unique challenge. Turnout patterns can vary dramatically, weather is a big factor and, as we saw in 2008, the timing of the caucuses and the momentum of the candidates themselves can influence the outcome. We took a look at four of the public pollsters that were tracking Iowa this go around.
Insider Advantage: The final two Insider Advantage polls ahead of the caucus had Paul (22 percent) and Romney (23 percent) running neck-and-neck. Their final poll days before the caucus showed Santorum (18 percent) passing Newt Gingrich (16 percent) for third place and Michele Bachmann (6 percent) a distant fifth.
The final survey polled 729 likely voters but was done in just one night, which is always a little iffy methodologically. It also showed just 3 percent undecided, which is low given the volatility leading up to the caucus. Looking at the trend from their polling, Insider Advantage had little relative movement between Romney, Paul, and Gingrich while showing a steady Santorum gain from Bachmann’s fall.
PPP: Public Policy Polling is a Democratic firm that exclusively uses auto or IVR polling. These polls can be very accurate in high turnout general elections, but how did they stack up given the low-turnout party caucus? The last PPP poll had Paul (20 percent) virtually tied with Romney (19 percent), with Santorum (18 percent) closing the gap quickly. Looking over PPP’s last few polls, they showed Paul fading slightly, Romney gaining and Santorum sprinting to the finish.
ARG: American Research Group has also been polling on a regular basis in Iowa. They showed Romney (22 percent) with a little bit of daylight between him and Paul (17 percent), Santorum (16 percent), and Gingrich (15 percent). ARG’s trend lines showed Romney treading water with Santorum gaining momentum, Paul ticking up slightly and Gingrich down a point.
Des Moines Register: The Des Moines Register has an excellent reputation for their Iowa caucus numbers, but they (smartly) don’t poll as often as some of the public pollsters we looked at, so the trend lines are spread much farther apart. That said, the final Register poll showed Romney in front (24 percent) followed closely by Paul (22 percent) and then a significant drop off to Santorum (15 percent) and Gingrich (12 percent).
With 100 percent of the vote in, Romney and Santorum each had 25 percent, Paul was third with 21 percent, Gingrich fourth at 13 percent, Perry at 10 percent and Bachmann at 5 percent. At the end of the day, if you combine the polling, the numbers proved to be solid predictors of the eventual outcome.
Romney was out front in most, but not gaining or losing support—his storyline for much of the past year—and every poll showed Santorum gaining steam. Paul had an average level of support of 20 percent over the four polls and ended with 21 percent support. It’s also clear that Gingrich, Perry and Bachmann supporters who defected landed with Santorum.
Going forward, there is considerably less polling being done in New Hampshire and South Carolina but enough to make it interesting to see how Iowa affects the numbers. Both states are also interesting to poll in for presidential primaries, especially because there isn’t a contested race on the Democratic side. We’ll see how the numbers there stack up. Stay tuned.
Stefan Hankin is founder and president of Lincoln Park Strategies, a Washington D.C.-based public opinion firm. Follow him on Twitter at @LPStrategies